From: Alvaro Herrera [mailto:alvherre@2ndquadrant.com]
> The more aggressive action is to backpatch 943576bddcb5 ("Make autovacuum
> more aggressive to remove orphaned temp tables") which is currently only
> in pg11. We would put the new PGPROC member at the end of the struct, to
> avoid ABI incompatibilities, but it'd bring trouble for extensions that
> put PGPROC in arrays. I checked the code of some known extensions; found
> that pglogical uses PGPROC, but only pointers to it, so it wouldn't be
> damaged by the proposed change AFAICS.
+1
I think this is a bug from a user's perspective that garbage is left. I want to believe that fixing bugs of PostgreSQL
itselfare prioritized over the ABI compatibility for extensions, if we have to choose one of them.
> Another possibly useful change is to make DISCARD ALL and DISCARD TEMP delete
> everything in what would be the backend's temp namespace, even if it hasn't
> been initialized yet. This would cover the case where a connection pooler
> keeps the connection open for a very long time, which I think is a common
> case.
That sounds good.
Regards
Takayuki Tsunakawa